Information Technology Reference
In-Depth Information
the ability to write Horn-like rules expressed in terms of OWL concepts to infer
new knowledge from existing OWL KB. For instance, each type predicate can be
expressed as a class. Each profile can be expressed as an OWL class whose instances
are the modules available to that profile (we exploit the multiple classification of SW
languages). Module type hierarchies can be expressed through
subclassOf
relation-
ships between the corresponding classes. All rules regarding performability and the
hierarchical organization of tasks can be expressed as SWRL rules.
In a DBMS-approach all facts can be stored in a relational database, while
Recursive SQL
can be used for expressing the rules. Specifically, each type pred-
icate can be expressed as a relational table with tuples the modules of that type.
Each profile can be expressed as an additional relational table, whose tuples will be
the modules known by that profile. All rules regarding task performability, hierar-
chical organisation of tasks, and the module type hierarchies, can be expressed as
Datalog queries. Recursion is required for being able to express the properties (e.g.
transitivity) of dependencies. Note that there are many commercial SQL servers that
support the SQL:1999 syntax regarding recursive SQL (e.g. Microsoft SQL Server
2005, Oracle 9i, and IBM DB2).
The following table (Table
8.1
) summarises the various implementation
approaches and describes how the elements of the model can be implemented. The
table does not contain any information about the Prolog approach since the terminol-
ogy we used for founding the model (Sect.
8.2.1.1
) is the same with that of Prolog.
Table 8.1
Implementation approaches for disjunctive dependencies
What
DB approach
Semantic web approach
Module type predicates
Relational table
Class
Facts regarding modules
(and their types)
Tuples
class instances
DC profile
Relational table
Class
DC profile contents
Tuples
class instances
Task predicates
IDB predicates
predicates appearing in rules
Task type hierarchy
Datalog rules, or
isa
if an
ORDBMS is used
subclassOf
Task performability
Datalog queries (recursive
SQL)
Rules
8.4 Summary
Although this Chapter is heavy going and the theory is not yet complete, it is worth
knowing that there is a theoretical basis for techniques to deal with Knowledge
Bases. Of particular importance is that some automation is possible in identifying
and fillings gaps in knowledge dependencies.
